Journal of Archaeological Science 34 (2007) 675e680

http://www.elsevier.com/locate/jas

Ancient DNA and the genetic signature of ancient Greek manuscripts

N. Poulakakis a,*, A. Tselikas b,c, I. Bitsakis d, M. Mylonas a,e, P. Lymberakis a
a
Natural History Museum of Crete, University of Crete, Knosos Av., P.O. Box 2208, 71409 Irakleio, Crete, Greece
b
National Bank of Greece Cultural Foundation (CF-NBG), P. Skouze st. 3, 10560 Athens, Greece
c
‘‘ARETHAS’’ Institute e Mediterranean Research Institut for Palaeography, Bibliology and History of Texts, 8 Tsituri str.,
Chalandri Athens 15231, Greece
d
K. Palaiologou 6, 16232 Vyron, Athens, Greece
e
Department of Biology, University of Crete, Vassilika Vouton, P.O. Box 2208, 71409 Irakleio, Crete, Greece
Received 4 January 2006; received in revised form 19 June 2006; accepted 20 June 2006

Abstract

Determination of the species origin of historic objects is one of the common tasks of ancient DNA (aDNA) analysis. DNA recovered from
archaeological and palaeontological remains allows going back in time and revealing the genetic signature of several human tools. Comparisons
of this signature with DNA sequence from recent animals (wild and domestic goats) from several Mediterranean regions are expected to allow us
to identify a geographical origin for the biological material used to produce the Greek parchment manuscripts. Here, we have realised an
experiment based on which it is possible to recover DNA from ancient parchment fragments (three Greek parchment manuscripts of relatively
recent age: 13th to 16th century AD). The analysis of the three Greek manuscripts has shown that most signature documents have goat-related
sequences (Capra spp.). As demonstrated, DNA of animals whose skins furnished the parchment pages of ancient and medieval books may
survive in that parchment, enabling not only to determine the species of the animal from which the skin had been taken, but moreover, it might
even be possible to reconstitute the history of the herds from which they originated.
Ó 2006 Elsevier Ltd. All rights reserved.

Keywords: Ancient DNA; Deamination; Greek manuscripts; Evolution

1. Introduction famed libraries of Alexandria, the Egyptians banned the export

Animal skin parchment is one of the oldest substances used
by human. Like rocks, sticks or other vegetable matter, people
had access to animals and tried to put every piece of an animal
to full use. Skins have been used for clothing, tools, and
weapons. In historical times, another form of animal skin,
parchment (pergamena in Latin), was an essential ‘‘vehicle’’
of cultural activity in Eurasia [2]. It is known that skins
were used as writing surfaces as early as 2700 BC in the IV
dynasty in Egypt. In an effort to stifle any possible challenge
to the amassing of knowledge that was taking place in the
* Corresponding author. Molecular Systematic Lab, Natural History Museum
of Crete, University of Crete, Knosou Av., P.O. Box 2208, 71409 Irakleio,
Crete, Greece. Tel./fax: þ30 2810 324 366.
E-mail address: poulakakis@nhmc.uoc.gr (N. Poulakakis).
of papyrus, which was the most common writing surface of the
time.
Parchment is a material for the pages of the codex, the
manuscript book, made from fine sheepskin, goatskin or calf-
skin. According to the historians, it was invented about the
beginning of the 2nd century (Before Christ, BC), in Perga-
mon, as a substitute for papyrus. Pergamon or Pergamum
(modern day Bergama in Turkey) was a Greek city, in north-
western Anatolia, 16 miles from the Aegean Sea, located on
a promontory on the north side of the river Caicus (modern
day Bakir), that became an important kingdom during the
Hellenistic period, under the Attalid dynasty, 282e129 BC.
Pergamon had a great library that rivaled the famous Library
of Alexandria. As prices rose for papyrus, while the reed was
overharvested towards local extinction in the Nile delta, Per-
gamon ‘‘adapted’’ by increasing use of parchment. Though

0305-4403/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2006.06.013
676 N. Poulakakis et al. / Journal of Archaeological Science 34 (2007) 675e680
the Assyrians and the Babylonians impressed their cuneiform
on clay tablets, they also wrote on parchment from the 6th
century BC onward [36]. The use of parchment for making
manuscript books expanded throughout Mediterranean, Asian
and African countries. Parchment was also the main material
of making manuscripts books until the 16th century AC. It
was used as well by the chancelleries of the ancient and me-
dieval states and the ecclesiastical authorities until 19th
century.
The ability to recover biomolecules, most importantly
DNA, from ancient remains has opened new research that
has many implications [13,24,34 and references therein]. Ac-
cess to ancient DNA (aDNA) provides the opportunity to study
the genetic material of past organisms and identify individual
and population histories. There are numerous studies in which
the retrieving of ancient DNA from post-mortem material pro-
vides scientists with powerful molecular evidence to examine
the genetic changes and phylogenetic relationships between
extant and extinct taxa, even when morphological and anatom-
ical information is limited.
The parchment on which most medieval manuscripts were
written was made from animal skin by removal of hair
and flesh, followed by drying and scraping [36]. When an
organism dies, its DNA normally becomes degraded by
endogenous nucleases. Under fortunate circumstances, such
as rapid desiccation, low temperatures or high salt concen-
trations, nucleases can themselves become destroyed or
inactivated before all nucleic acids are reduced to mononucle-
otides [13]. The treatments used in this process were fairly
mild, and there is a good chance that some of the DNA in
the skin would have been left intact. But will it have survived
through time? Past investigations have shown that ancient
DNA can be detected not only in various palaeontological, ar-
chaeological and museum materials [13,24,34 and references
therein], including parchments [36,3,30]. The post-mortem
instability of nucleic acids is central to the methodological
problems inherent in aDNA research. Early research showed
that post-mortem DNA decay was characterized by strand
breaks, baseless sites, miscoding lesions and crosslinks, and
that these caused both sequencing artefacts and the preferen-
tial amplification of undamaged contaminant DNA
[15,22,23,24,34]. However, the relative rates of various types
of DNA damage and their mode of accumulation remain
poorly characterized [6,7].
Numerous parchments (more than 3000 manuscripts) are
available in several Greek libraries (Mount Athos, Athens,
Patmos, Jerusalem, Alexandria, Sinai, etc.) about the origins
of which we have limited knowledge. A number of questions
concerning the origin and production of the Greek parchments
may be addressed using DNA analysis. Based on radiocarbon
and other analyses, these manuscripts date between the 8th and
16th century and were for the most part written on what is
thought to be goat or sheepskin parchment. Here, we tried in
collaboration with the Cultural Foundation of National Bank
of Greece (CF-NBG) that has the opportunity to have access
to these manuscripts, to reveal the genetic signature of the
Greek manuscript.
2. Materials and methods
2.1. DNA extraction and amplification
Standard ancient DNA handling conditions and dedicated
equipment were used for parchment collection, cleaning and
DNA extractions [17,21]. Equipment and reagents were dedi-
cated solely for ancient DNA work and extractions, and ampli-
fications were carried out in a laboratory where no mammalian
DNA except human had been previously used. Disposable
equipment was used whenever possible, and reusable equip-
ment was soaked in 0.5% sodium hypochlorite and then ex-
posed to UV light for 1 h prior to use.
Genomic DNA was extracted from six parchments (Hel-
lenic manuscripts), the age of which ranged between 13th and
16th century AD, using silica-based and phenolechloroform
extraction methods, but the phenolechloroform method was
more. The following protocol was used for phenolechloroform
DNA isolation. DNA was extracted from small pieces
(approximately 0.5 cm2) of parchment in a solution containing
50 mM TriseHCl (pH 8.0), 100 mM NaCl, 10e25 mM
EDTA, 2% SDS, and at least 200 mg/ml proteinase K for
16e24 h at 56  C, followed by phenolechloroform extraction.
DNA purification was completed by precipitation with isopro-
panol or by concentrating with Centricon YM centrifugal filter
devices (with Amicon filter).
Independent extracts from the same sample or samples
from different parchments were used as template for PCR am-
plification. PCR was performed using AmpliTaq Gold (Per-
kineElmer, USA) with a MgCl2 concentration of 3 mM,
following the supplier’s instructions, except for the addition
of bovine serum albumin to a final concentration of 0.25 mg/
ml. Two pairs of primer were used to assemble a contiguous
sequence of up to 320 bp from the cytochrome b gene (cyt
b). Primers were designed to target overlapping DNA frag-
ments of 150e200 bp (Fig. 1A). Amplifications were done
in a M.J. Thermocycler with a 5 min activation step at
94  C, followed by 60 cycles of 94  C for 30 s, 42  C for
60 s and 72  C for 30 s. PCR products were isolated from
3% agarose gels and melted in 50 ml ddH2O, of which 5 ml
were used for reamplification for 30 cycles under the PCR
conditions described, except that the activation step was pro-
longed to 7 min and the annealing temperature was 45  C.
At least two sterile negative controls were used for each
reaction to detect contamination throughout the extraction
and amplification reaction. For each sample studied the entire
analysis was undertaken twice to confirm the sequences
obtained.
2.2. Cloning and sequencing
Reamplification products product were then directly ligated
overnight at 4  C into the pGEM (Promega) plasmid, followed
by transformation in Escherichia coli JM109 competent cells
(Promega). Positive colonies were checked for correctly sized
inserts using a PCR assay to amplify the region between the
vector’s SP6 and T7 promoter sites. PCR products of the
 N. Poulakakis et al. / Journal of Archaeological Science 34 (2007) 675e680 677

Fig. 1. (A) Position of the targeted fragments of mitochondrial cyt b gene, the primers and the produced sequence: schematic map. The length of each fragment is
indicated in base pairs. The nucleotide numbering is based on the sequence of Capra hircus (GenBank AB004072). (B) Alignment of the mitochondrial DNA
clones sequenced from three amplifications products for each one the three ‘‘successful’’ parchments used in this study. The consensus sequence was generated
by Consensus Confidence Program (v.1.12). The high percentage of C/G/TA changes is probably due to deamination of C residues in the ancient DNA templates.
Numbers above the sequences indicates consistent, singleton and other substitutions among DNA sequences.

correct size from this assay were used as templates for auto-
mated cycle sequencing on an ABI 377 (Applied Biosystems),
according to manufacturer’s instructions. GenBank accession
numbers for the sequences obtained are DQ778621e
DQ778623.
2.3. Phylogenetic analysis
Sequences were aligned using the ClustalX program pack-
age [27]. The MEGA computer package (v.2) [29] was used to
determine the number and type of nucleotide substitutions in
678 N. Poulakakis et al. / Journal of Archaeological Science 34 (2007) 675e680

pair-wise comparisons of sequences, to measure the degree of
divergence between sequences and to identify the unique se-
quences for phylogenetic analysis. For phylogenetic analysis,
a data set of 24 cyt b sequences was used (including the three
parchment sequences of this study). Bayesian inference
[10,35] analysis and Maximum Likelihood (ML) [4] were per-
formed with the MrBayes (v3.1B) [10] and PAUP*4.0b10 [8],
respectively, using a model of substitution [HKY þ G model
[11] with Ti/Tv ¼ 11.8563, a ¼ 0.091 and A ¼ 23.1, C ¼ 24.7,
G ¼ 18, T ¼ 34.2] that was chosen by performing hierarchical
likelihood-ratio tests [14] in Modeltest (v.3.06) [1]. For BI,
the analysis was run with four chains for 107 generations
and the current tree was saved to file every 100 generations.
This generated an output of 105 trees. The elnL stabilized
after approximately 104 generations and the first 103 trees
(10% ‘‘burn-in’’ in Bayesian terms, chain had not become
stationary) were discarded as a conservative measure to avoid
the possibility of including random, sub-optimal trees. The
percentage of samples recovering any particular clade in
a BI analysis represents that clade’s posterior probability
[35]. We used one of the methods of [19] to assure that
our analyses were not trapped on local optima. In particular,
the posterior probabilities for individual clades obtained from
separate analyses (four runs) were compared for congruence
[28], given the possibility that the analyses could appear to
converge on the same ln-likelihood value while actually sup-
porting incongruent phylogenetic trees.
3. Results and discussion
in a specimen, some extracts will fail to contain DNA mole-
cules by chance [13]. Consequently, it was an expected result
the fact that endogenous DNA was amplified from three out of
six specimens of ancient Greek parchment remains used in this
study (Table 1).
While most post-mortem DNA damage events fragment the
molecule and prevent it from being amplified, a small propor-
tion merely generates miscoding lesions [34]. These are man-
ifested as base modifications in the amplified sequence,
changing the appearance of aDNA template. The few detailed
studies of miscoding lesions concur with earlier hypotheses
[e.g., see 23,22,15] that the majority of changes arise from
the deamination of C to uracil (U), an analogue of T, or the
deamination of A to hypoxanthine (HX), an analogue of G
[6,7,26,5,32]. Ancient DNA template is a mixture of mole-
cules comprising varying amounts of the correct endogenous
sequence, damaged endogenous sequence, contaminant se-
quence and damaged contaminant sequence. Therefore, PCR
products from aDNA template are also likely to be a mixture
of misamplified damaged template and contaminant template.
In order to determine the nature of the DNA sequences am-
plified, the two overlapping cyt b fragments (194 and 150 bp,
Fig. 1A) were cloned and the inserts of 22e28 clones were se-
quenced (Fig. 1B) for the three ‘‘successful’’ specimens. To
obtain the true aDNA sequence, it is not sufficient to generate
a single direct sequence of the mixture, even where the authen-
tic aDNA is the most abundant in the component mixture.
Only bacterial cloning can elucidate if the sequence is real
or not. Bower et al. [12] calculated that the whole number

When an organism dies, its DNA normally becomes de-

graded to small average size, generally between 100 and Table 1
List of the specimens of Bovidae used in molecular analyses (species name,
500 bp [13,23] by both enzymatic and nonenzymatic (hydro-
number of specimens, samples age, and GenBank accession numbers of se-
lytic and oxidotic) processes [9,22]. Nevertheless, DNA has quence data)
been recovered from archaeological and palaeontological
Specimens Samples Age GenBank
remains [13,24,34], as long as several fortunate environmental accession No.
conditions, such as rapid desiccation, low temperatures
Parchment 1 (Evagelistario) 1 13th century AD DQ778621
or high salt concentrations, may prolong DNA survival Parchment 2 (Pateriko) 1 13th century AD DQ778622
[13,24,34,22,31,33]. Parchment 3 (Galliko) 1 16th century AD DQ778623
However, the field of ancient DNA is fraught with technical Capra hircus 6 Recent AF217254,
pitfalls and needs stringent criteria to ensure the reliability of X56289,
AB004072,
results. Quite a few review articles reported the basic authen-
D84201,
ticity criteria for the determination of ancient DNA sequences AB004075,
[13,24,34,20,18,26]. Critical steps such as extraction and PCR AJ231402
controls, inverse correlation between amplicon length and am- Capra ibex 1 Recent AF034735
plification efficiency, amplification from a second extract, Capra pyrenaica 2 Recent AJ010048,
AJ213406
cloning, phylogenetic sense are widely accepted. In addition
Capra falconeri 1 Recent AF034736
when an unexpected result is obtained, the reproduction in Capra caucasica 1 Recent AF034738
a second laboratory is necessary [13]. Capra cylindricornis 1 Recent AF034737
Given that authentic ancient DNA is typically highly de- Capra aegagrus 1 Recent AF034739,
graded, shorter target DNA fragments should be sought. For AJ231408
Capra nubiana 1 Recent AF034740
this reason, we designed two pairs of primers (Fig. 1A) based
Capra sibirica 1 Recent AF034734
on previously published caprinae sequences that amplify short, Ovis aries 1 Recent AF034730
nonhuman, mtDNA fragments (194 and 150 bp, respectively). Ovis ammon 1 Recent AF242350
For each extraction (30 in total, five for each specimen), Bos Taurus 1 Recent NC001567
amplifications were performed using these specific pairs of Suc scrofa 1 Recent AJ314558
Homo sapiens 1 Recent AY255180
primer. It is known that if low amounts of DNA are preserved
 N. Poulakakis et al. / Journal of Archaeological Science 34 (2007) 675e680 679

of clones that need to be sampled in order to be 95% confident goat, or calfskin were used. However, the morphological spe-
of identifying the most abundant sequence present at 70% in cies determination was impossible. The sequences of parch-
the ancient sample are 20. ments are 323 bp and correspond to positions 64e386 of
We generated the consensus sequence using the free-access Capra hircus sequence (GenBank AB004072, Fig. 1A). Dur-
web-based program (Consensus Confidence Program, v.1.12, ing the phylogenetic analyses (Maximum likelihood and
http://www.mcdonald.cam.ac.uk), which constructs the most Bayesian Inference, see Section 2) identical topologies were
reliable consensus sequence from the user’s input clone se- recovered (Fig. 2). The parchment sequences were signifi-
quences and analyses the confidence limits for each nucleotide cantly different from the human sequence, demonstrating
position and for the whole consensus sequence. that the parchment material was not contaminated by human
In total, 75 parchment clones for each fragment (194 and DNA either in the handling of the parchment during collection
150 bp) from nine initial PCR reactions (three for each ‘‘suc- or during the laboratory manipulations and were most closely
cessful’’ parchment) were analyzed. When we scrutinized the related to that of goats (C. hircus), giving us the opportunity to
nucleotide substitutions observed among the clones (Fig. 1B), establish the genetic signature unique for each parchment. To
we found that for two amplification products (PCR 1 in posi- further test all the possible competing monophyly hypotheses,
tion 241 of parchment 1 and PCR 1 in position 63 of parch- we used the ShimodairaeHasegawa test [25] to compare the
ment 2, Fig. 1B) all clones that had been sequenced carried ML tree topology with the alternative ones (forcing the parch-
substitutions that distinguished them from all the clones of ment sequences within other Bovidae species). The SH test in
other independent amplifications from the same DNA extract. all cases was in favor ( p < 0.0001) of the topology presented
Hofreiter et al. [32] stated that substitutions (consistent substi- in Fig. 2.
tutions) that occurred consistently among clones within one As demonstrated, DNA of animals whose skins furnished
amplification, but were not reproducible in other amplifica- the parchment pages of ancient and medieval books may sur-
tions are likely to be the result of nucleotide misincorporations vive in that parchment. It might be possible not only to deter-
that occurred during the first cycle of the PCR in cases where mine the species of the animal from which the skin had been
the amplification started from a single DNA molecule and they taken, but moreover, it might even be possible to reconstitute
represent errors induced during replication of the ancient DNA the history of the herds from which they originated. In
molecules [20,32,16]. Moreover, one singleton substitution a broader perspective, aDNA obtained from the parchment
was observed in a single clone (PCR 1, clone 5, position fragments may help answer some interesting questions as:
188 in the sequence of parchment 3, Fig. 1B), which probably what is the origin of the manuscripts in the Greek libraries col-
is due to Thermus aquaticus (Taq) DNA polymerase misincor- lections? Which items can be grouped together as originating
poration in later cycles of the PCR [32] and is rather an error
that occurs during replication of newly synthesised molecules
than a modification present in the original DNA molecules ex-
tracted from the parchment. Finally, for the rest PCR amplifi-
cations (four PCRs and 59 clones), where no consistent or
singleton substitutions were observed, four other substitutions
that we found in a few clones (Fig. 1B), could be explained
either by DNA damage present in the ancient DNA template
or by subsequent PCR errors during amplification [32].
It is noteworthy that among all substitutions (two consis-
tent, one singleton substitution, and for other substitutions),
five were C/T substitutions and two were G/A substitu-
tions, showing an extreme bias (G/C/A/T changes). This
phenomenon, which is quite common in the DNA extracted
from archaeological and palaeontological samples, could be
explained by cytosine deamination and/or by the tendency of
Taq polymerase to add deoxyadenosine residues when it rea-
ches the ends of templates [6,7,26,32].
To ensure the authenticity of the ancient DNA sequences
[24,34] we repeated the extractions, the amplifications, the
cloning and the sequencing of the amplifiers (see Section 2).
The consensus sequences of the three parchments used for
a preliminary phylogenetic analysis with several Bovidae spe-
cies (Capra spp., Ovis spp., and Bos Taurus, see Table 1). Phy-
logenetic analysis is perhaps the final criterion for the Fig. 2. Cyt b phylogeny of several Bovidae sequences, including the produced
sequences of the three parchments of this study. Phylogenetic analyses of max-
authenticity of ancient DNA sequence [24,34]. We suspected imum likelihood (ML) and Bayesian inference (BI) produced trees with the
the origin of the DNA extracted from the three legal Greek same topology with regard to the major lineages. Only the BI tree is presented.
manuscripts (parchments), since for their construction sheep, Numbers indicate posterior probabilities values of BI.
680 N. Poulakakis et al. / Journal of Archaeological Science 34 (2007) 675e680
from the same or closely related parchments, craftsmen and/or
localities? Did different scribes use parchment originating
from various sources for their manuscripts, or did some have
specific providers? Does the collection contain manuscripts
from a single locality, or is it a collection representing contri-
butions from a wide region? What is the temporal element of
these questions? From another point of view, what are the phy-
logeographic relationships of contemporary domestic animals
with the stocks bred several hundreds years ago?
This study is the beginning of a hopefully fruitful project
that will continue over the next few years. We hope that the
analysis of DNA from parchment fragments will add a new
level of critical analysis to scroll knowledge, given the avail-
ability of more than 3000 manuscripts from several libraries
of Greece (Mount Athos, Athens, Patmos, Jerusalem, Alexan-
dria, Sinai).
References
[1] M.A. Bower, M. Spencer, S. Matsumura, R. Ellen, R. Nisbet, C.J. Howe,
How many clones need to be sequenced from a single forensic or ancient
DNA sample in order to determine a reliable consensus sequence? Nu-
cleic Acids Res. 33 (2005) 2549e2556.
[2] J. Burger, S. Hummel, B. Herrmann, W. Henke, DNA preservation. A mi-
crosatellite DNA study on ancient skeletal remains, Electrophoresis 20
(1999) 1722e1728.
[3] J. Burger, I. Pfeiffer, S. Hummel, R. Fuchs, B. Brenig, B. Herrmann,
Mitochondrial and nuclear DNA from (pre)historic hide-derived material,
Anc. Biomol. 3 (2001) 227e238.
[4] A. Cooper, H.N. Poinar, Ancient DNA: do it right or not at all, Science
289 (2000) 1139.
[5] J. Felsenstein, Evolutionary trees from DNA sequences: a maximum like-
lihood approach, J. Mol. Evol. 17 (1981) 368e376.
[6] M.T.P. Gilbert, E. Willerslev, A.J. Hansen, I. Barnes, L. Rudbeck,
N. Lynnerup, A. Cooper, Distribution patterns of postmortem damage
in human mitochondrial DNA, Am. J. Hum. Genet. 72 (2003) 32e47.
[7] M.T.P. Gilbert, A.J. Hansen, E. Willerslev, L. Rudbeck, I. Barnes,
N. Lynnerup, A. Cooper, Characterization of genetic miscoding lesions
caused by postmortem damage, Am. J. Hum. Genet. 72 (2003) 48e61.
[8] M.P.T. Gilbert, H.-J. Bandelt, M. Hofreiter, I. Barnes, Assessing ancient
DNA studies, Trends Ecol. Evol. 20 (2005) 541e544.
[9] O. Handt, M. Höss, M. Krings, S. Pääbo, Ancient DNA: methodological
challenges, Experientia 50 (1994) 524e529.
[10] O. Handt, M. Krings, R.H. Ward, S. Pääbo, The retrieval of ancient hu-
man DNA sequences, Am. J. Hum. Genet. 59 (1996) 376e386.
[11] A.J. Hansen, E. Willerslev, C. Wiuf, T. Mourier, P. Arctander, Statistical
evidence for miscoding lesions in ancient DNA templates, Mol. Biol.
Evol. 18 (2001) 262e265.
[12] M. Hasegawa, H. Kishino, T. Yano, Dating of the human-ape splitting by
a molecular clock of mitochondrial DNA, J. Mol. Evol. 22 (1985) 160e174.
[13] M. Hofreiter, D. Serre, H.N. Poinar, M. Kuch, S. Pääbo, Ancient DNA,
Nat. Rev. Genet. 2 (2001) 353e359.
[14] M. Hofreiter, V. Jaenicke, S. Serre, A. von Haeseler, S. Pääbo, DNA se-
quences from multiple amplifications reveal artifacts induced by cytosine
deamination in ancient DNA, Nucleic Acids Res. 29 (2001) 4793e4799.
[15] M. Höss, A. Dilling, A. Currant, S. Pääbo, Molecular phylogeny of the
extinct groung sloth Mylodon darwinii, Proc. Natl. Acad. Sci. U.S.A.
93 (1996) 181e185.
[16] J.P. Huelsenbeck, K.A. Crandall, Phylogeny estimation and hypothesis
testing using maximum likelihood, Ann. Rev. Ecol. Syst. 28 (1997)
437e466.
[17] J.P. Huelsenbeck, N.S. Imennov, Geographic origin of human mitochon-
drial DNA: accommodating phylogenetic uncertainty and model compar-
ison, Syst. Biol. 51 (1) (2002) 155e165.
[18] J.P. Huelsenbeck, F.P. Ronquist, MRBAYES: bayesian inference of phy-
logenetic trees, Bioinformatics 17 (2001) 754e755.
[19] M. Krings, A. Stone, R.W. Schmitz, H. Krainitzki, M. Stoneking,
S. Pääbo, Neandertal DNA sequences and the origin of modern humans,
Cell 90 (1997) 19e30.
[20] S. Kumar, K. Tamura, I.B. Jakobsen, M. Nei, MEGA2: molecular evolu-
tionary genetics analysis software, Bioinformatics 17 (2001) 1244e1245.
[21] D. Leaché, T.W. Reeder, Molecular systematics of the eastern fence liz-
ard (Sceloporus undulates): a comparison of parsimony, likelihood and
Bayesian approaches, Syst. Biol. 51 (2002) 44e68.
[22] T. Lindahl, Instability and decay of the primary structure of DNA, Nature
362 (1993) 709e715.
[23] S. Pääbo, Ancient DNA: extraction, characterization, molecular cloning,
and enzymatic amplification, Proc. Natl. Acad. Sci. U.S.A. 86 (1989)
1939e1943.
[24] S. Pääbo, H. Poinar, D. Serre, V. Jaenicke-Despres, J. Hebler,
N. Rohland, M. Kuch, J. Krause, L. Vigilant, M. Hofreiter, Genetic anal-
yses from ancient DNA, Annu. Rev. Genet. 38 (2004) 645e679.
[25] D. Posada, K.A. Crandall, Modeltest: testing the model of DNA substi-
tution, Bioinformatics 14 (1998) 817e818.
[26] F. Ronquist, J.P. Huelsenbeck, MRBAYES 3: Bayesian phylogenetic in-
ference under mixed models, Bioinformatics 19 (2003) 1572e1574.
[27] R. Shapiro, Damage to DNA caused by hydrolysis, in: E. Seeberg,
K. Kleppe (Eds.), Chromosome Damage and Repair, Plenum, New
York, 1981, pp. 3e12.
[28] H. Shimodaira, M. Hasegawa, Multiple comparisons of log-likelihoods
with applications to phylogenetic inference, Mol. Biol. Evol. 16 (1999)
1114e1116.
[29] C.I. Smith, A.T. Chamberlein, M.S. Riley, A. Cooper, C.B. Stringer,
M.J. Collins, Neanderthal DNA: not just old but old and cold? Nature
10 (2001) 771e772.
[30] M. Spencer, C.J. Howe, Authenticity of ancient-DNA results: a statistical
approach, Am. J. Hum. Genet. 75 (2004) 240e250.
[31] M. Stoneking, Ancient DNA: how do you know when you have it and
what can you do with it? Am. J. Hum. Genet. 57 (1995) 1259e1262.
[32] D.L. Swofford, PAUP 4.0b10a: Phylogenetic Analysis Using Parsimony
(and Other Methods), Sinauer, Sunderland, MA, 2002.
[33] J.D. Thompson, T.J. Gibson, F. Plewniak, F. Jeanmougin, D.G. Higgins,
The ClustalX windows interface: flexible strategies for multiple sequence
alignment aided by quality analysis tool, Nucleic Acids Res. 24 (1997)
4876e4882.
[34] E. Willerslev, A. Cooper, Ancient DNA, Proc. R. Soc. Lond. B 272
(2005) 3e16.
[35] E. Willerslev, A.J. Hansen, H.N. Poinar, Isolation of nucleic acids and
cultures from ice and permafrost, Trends Ecol. Evol. 19 (2004) 141e147.
[36] S.R. Woodward, G. Kahila, P. Smith, C. Greenblatt, J. Zias, M. Broshi,
Analysis of parchment fragments from the Judea Desert using DNA
techniques, in: D.W. Parry, S. Ricks (Eds.), Current Research and Tech-
nological Developments on the Dead Sea Scrolls, Leiden, 1996, pp.
215e238.